languages/LanguageConverter.php

   1 <?php
   2
   3 /**
   4  * Contains the LanguageConverter class and ConverterRule class
   5  * @ingroup Language
   6  *
   7  * @license http://www.gnu.org/copyleft/gpl.html GNU General Public License
   8  * @file
   9  */
  10
  11 /**
  12  * base class for language convert
  13  * @ingroup Language
  14  *
  15  * @author Zhengzhu Feng <zhengzhu@gmail.com>
  16  * @maintainers fdcn <fdcn64@gmail.com>, shinjiman <shinjiman@gmail.com>
  17  */
  18 class LanguageConverter {
  19         var $mPreferredVariant='';
  20         var $mMainLanguageCode;
  21         var $mVariants, $mVariantFallbacks, $mVariantNames;
  22         var $mTablesLoaded = false;
  23         var $mTables;
  24         var $mTitleDisplay='';
  25         var $mDoTitleConvert=true, $mDoContentConvert=true;
  26         var $mManualLevel; // 'bidirectional' 'unidirectional' 'disable' for each variants
  27         var $mTitleFromFlag = false;
  28         var $mCacheKey;
  29         var $mLangObj;
  30         var $mMarkup;
  31         var $mFlags;
  32         var $mDescCodeSep = ':',$mDescVarSep = ';';
  33         var $mUcfirst = false;
  34
  35         const CACHE_VERSION_KEY = 'VERSION 6';
  36
  37         /**
  38          * Constructor
  39          *
  40          * @param string $maincode the main language code of this language
  41          * @param array $variants the supported variants of this language
  42          * @param array $variantfallback the fallback language of each variant
  43          * @param array $markup array defining the markup used for manual conversion
  44          * @param array $flags array defining the custom strings that maps to the flags
  45          * @param array $manualLevel limit for supported variants
  46          * @public
  47          */
  48         function __construct($langobj, $maincode,
  49                                                                 $variants=array(),
  50                                                                 $variantfallbacks=array(),
  51                                                                 $markup=array(),
  52                                                                 $flags = array(),
  53                                                                 $manualLevel = array() ) {
  54                 $this->mLangObj = $langobj;
  55                 $this->mMainLanguageCode = $maincode;
  56                 $this->mVariants = $variants;
  57                 $this->mVariantFallbacks = $variantfallbacks;
  58                 global $wgLanguageNames;
  59                 $this->mVariantNames = $wgLanguageNames;
  60                 $this->mCacheKey = wfMemcKey( 'conversiontables', $maincode );
  61                 $m = array(
  62                         'begin'=>'-{',
  63                         'flagsep'=>'|',
  64                         'unidsep'=>'=>', //for unidirectional conversion
  65                         'codesep'=>':',
  66                         'varsep'=>';',
  67                         'end'=>'}-'
  68                 );
  69                 $this->mMarkup = array_merge($m, $markup);
  70                 $f = array(
  71                         // 'S' show converted text
  72                         // '+' add rules for alltext
  73                         // 'E' the gave flags is error
  74                         // these flags above are reserved for program
  75                         'A'=>'A',       // add rule for convert code (all text convert)
  76                         'T'=>'T',       // title convert
  77                         'R'=>'R',       // raw content
  78                         'D'=>'D',       // convert description (subclass implement)
  79                         '-'=>'-',       // remove convert (not implement)
  80                         'H'=>'H',       // add rule for convert code (but no display in placed code )
  81                         'N'=>'N'        // current variant name
  82                 );
  83                 $this->mFlags = array_merge($f, $flags);
  84                 foreach( $this->mVariants as $v)
  85                         $this->mManualLevel[$v]=array_key_exists($v,$manualLevel)
  86                                                                 ?$manualLevel[$v]
  87                                                                 :'bidirectional';
  88         }
  89
  90         /**
  91          * @public
  92          */
  93         function getVariants() {
  94                 return $this->mVariants;
  95         }
  96
  97         /**
  98          * in case some variant is not defined in the markup, we need
  99          * to have some fallback. for example, in zh, normally people
 100          * will define zh-hans and zh-hant, but less so for zh-sg or zh-hk.
 101          * when zh-sg is preferred but not defined, we will pick zh-hans
 102          * in this case. right now this is only used by zh.
 103          *
 104          * @param string $v the language code of the variant
 105          * @return string array the code of the fallback language or false if there is no fallback
 106          * @public
 107          */
 108         function getVariantFallbacks($v) {
 109                 if( isset( $this->mVariantFallbacks[$v] ) ) {
 110                         return $this->mVariantFallbacks[$v];
 111                 }
 112                 return $this->mMainLanguageCode;
 113         }
 114
 115         /**
 116          * get preferred language variants.
 117          * @param boolean $fromUser Get it from $wgUser's preferences
 118          * @return string the preferred language code
 119          * @public
 120          */
 121         function getPreferredVariant( $fromUser = true ) {
 122                 global $wgUser, $wgRequest, $wgVariantArticlePath, $wgDefaultLanguageVariant;
 123
 124                 if($this->mPreferredVariant)
 125                         return $this->mPreferredVariant;
 126
 127                 // see if the preference is set in the request
 128                 $req = $wgRequest->getText( 'variant' );
 129                 if( in_array( $req, $this->mVariants ) ) {
 130                         $this->mPreferredVariant = $req;
 131                         return $req;
 132                 }
 133
 134                 // check the syntax /code/ArticleTitle
 135                 if($wgVariantArticlePath!=false && isset($_SERVER['SCRIPT_NAME'])){
 136                         // Note: SCRIPT_NAME probably won't hold the correct value if PHP is run as CGI
 137                         // (it will hold path to php.cgi binary), and might not exist on some very old PHP installations
 138                         $scriptBase = basename( $_SERVER['SCRIPT_NAME'] );
 139                         if(in_array($scriptBase,$this->mVariants)){
 140                                 $this->mPreferredVariant = $scriptBase;
 141                                 return $this->mPreferredVariant;
 142                         }
 143                 }
 144
 145                 // get language variant preference from logged in users
 146                 // Don't call this on stub objects because that causes infinite
 147                 // recursion during initialisation
 148                 if( $fromUser && $wgUser->isLoggedIn() )  {
 149                         $this->mPreferredVariant = $wgUser->getOption('variant');
 150                         return $this->mPreferredVariant;
 151                 }
 152
 153                 // see if default variant is globaly set
 154                 if($wgDefaultLanguageVariant != false  &&  in_array( $wgDefaultLanguageVariant, $this->mVariants )){
 155                         $this->mPreferredVariant = $wgDefaultLanguageVariant;
 156                         return $this->mPreferredVariant;
 157                 }
 158
 159                 # FIXME rewrite code for parsing http header. The current code
 160                 # is written specific for detecting zh- variants
 161                 if( !$this->mPreferredVariant ) {
 162                         // see if some supported language variant is set in the
 163                         // http header, but we don't set the mPreferredVariant
 164                         // variable in case this is called before the user's
 165                         // preference is loaded
 166                         $pv=$this->mMainLanguageCode;
 167                         if(array_key_exists('HTTP_ACCEPT_LANGUAGE', $_SERVER)) {
 168                                 $header = str_replace( '_', '-', strtolower($_SERVER["HTTP_ACCEPT_LANGUAGE"]));
 169                                 $zh = strstr($header, $pv.'-');
 170                                 if($zh) {
 171                                         $pv = substr($zh,0,5);
 172                                 }
 173                         }
 174                         // don't try to return bad variant
 175                         if(in_array( $pv, $this->mVariants ))
 176                                 return $pv;
 177                 }
 178
 179                 return $this->mMainLanguageCode;
 180
 181         }
 182
 183         /**
 184          * dictionary-based conversion
 185          *
 186          * @param string $text the text to be converted
 187          * @param string $toVariant the target language code
 188          * @return string the converted text
 189          * @private
 190          */
 191         function autoConvert($text, $toVariant=false) {
 192                 $fname="LanguageConverter::autoConvert";
 193
 194                 wfProfileIn( $fname );
 195
 196                 if(!$this->mTablesLoaded)
 197                         $this->loadTables();
 198
 199                 if(!$toVariant)
 200                         $toVariant = $this->getPreferredVariant();
 201                 if(!in_array($toVariant, $this->mVariants))
 202                         return $text;
 203
 204                 /* we convert everything except:
 205                    1. html markups (anything between < and >)
 206                    2. html entities
 207                    3. place holders created by the parser
 208                 */
 209                 global $wgParser;
 210                 if (isset($wgParser) && $wgParser->UniqPrefix()!=''){
 211                         $marker = '|' . $wgParser->UniqPrefix() . '[\-a-zA-Z0-9]+';
 212                 } else
 213                         $marker = "";
 214
 215                 // this one is needed when the text is inside an html markup
 216                 $htmlfix = '|<[^>]+$|^[^<>]*>';
 217
 218                 // disable convert to variants between <code></code> tags
 219                 $codefix = '<code>.+?<\/code>|';
 220                 // disable convertsion of <script type="text/javascript"> ... </script>
 221                 $scriptfix = '<script.*?>.*?<\/script>|';
 222                 // disable conversion of <pre xxxx> ... </pre>
 223                 $prefix = '<pre.*?>.*?<\/pre>|';
 224
 225                 $reg = '/'.$codefix . $scriptfix . $prefix . '<[^>]+>|&[a-zA-Z#][a-z0-9]+;' . $marker . $htmlfix . '/s';
 226
 227                 $matches = preg_split($reg, $text, -1, PREG_SPLIT_OFFSET_CAPTURE);
 228
 229                 $m = array_shift($matches);
 230
 231                 $ret = $this->translate($m[0], $toVariant);
 232                 $mstart = $m[1]+strlen($m[0]);
 233                 foreach($matches as $m) {
 234                         $ret .= substr($text, $mstart, $m[1]-$mstart);
 235                         $ret .= $this->translate($m[0], $toVariant);
 236                         $mstart = $m[1] + strlen($m[0]);
 237                 }
 238                 wfProfileOut( $fname );
 239                 return $ret;
 240         }
 241
 242         /**
 243          * Translate a string to a variant
 244          * Doesn't process markup or do any of that other stuff, for that use convert()
 245          *
 246          * @param string $text Text to convert
 247          * @param string $variant Variant language code
 248          * @return string Translated text
 249          * @private
 250          */
 251         function translate( $text, $variant ) {
 252                 wfProfileIn( __METHOD__ );
 253                 if( !$this->mTablesLoaded )
 254                         $this->loadTables();
 255                 $text = $this->mTables[$variant]->replace( $text );
 256                 wfProfileOut( __METHOD__ );
 257                 return $text;
 258         }
 259
 260         /**
 261          * convert text to all supported variants
 262          *
 263          * @param string $text the text to be converted
 264          * @return array of string
 265          * @public
 266          */
 267         function autoConvertToAllVariants($text) {
 268                 $fname="LanguageConverter::autoConvertToAllVariants";
 269                 wfProfileIn( $fname );
 270                 if( !$this->mTablesLoaded )
 271                         $this->loadTables();
 272
 273                 $ret = array();
 274                 foreach($this->mVariants as $variant) {
 275                         $ret[$variant] = $this->translate($text, $variant);
 276                 }
 277
 278                 wfProfileOut( $fname );
 279                 return $ret;
 280         }
 281
 282         /**
 283          * convert link text to all supported variants
 284          *
 285          * @param string $text the text to be converted
 286          * @return array of string
 287          * @public
 288          */
 289         function convertLinkToAllVariants($text) {
 290                 if( !$this->mTablesLoaded )
 291                         $this->loadTables();
 292
 293                 $ret = array();
 294                 $tarray = explode($this->mMarkup['begin'], $text);
 295                 $tfirst = array_shift($tarray);
 296
 297                 foreach($this->mVariants as $variant)
 298                         $ret[$variant] = $this->translate($tfirst,$variant);
 299
 300                 foreach($tarray as $txt) {
 301                         $marked = explode($this->mMarkup['end'], $txt, 2);
 302
 303                         foreach($this->mVariants as $variant){
 304                                 $ret[$variant] .= $this->mMarkup['begin'].$marked[0].$this->mMarkup['end'];
 305                                 if(array_key_exists(1, $marked))
 306                                         $ret[$variant] .= $this->translate($marked[1],$variant);
 307                         }
 308
 309                 }
 310
 311                 return $ret;
 312         }
 313
 314
 315         /**
 316          * apply manual conversion
 317          * @private
 318          */
 319         function applyManualConv($convRule){
 320                 // use syntax -{T|zh:TitleZh;zh-tw:TitleTw}- for custom conversion in title
 321                 $title = $convRule->getTitle();
 322                 if($title){
 323                         $this->mTitleFromFlag = true;
 324                         $this->mTitleDisplay =  $title;
 325                 }
 326
 327                 //apply manual conversion table to global table
 328                 $convTable = $convRule->getConvTable();
 329                 $action = $convRule->getRulesAction();
 330                 foreach($convTable as $v=>$t) {
 331                         if( !in_array($v,$this->mVariants) )continue;
 332                         if( $action=="add" )
 333                                 $this->mTables[$v]->mergeArray($t);
 334                         elseif ( $action=="remove" )
 335                                 $this->mTables[$v]->removeArray($t);
 336                 }
 337         }
 338
 339         /**
 340          * Convert text using a parser object for context
 341          * @public
 342          */
 343         function parserConvert( $text, &$parser ) {
 344                 global $wgDisableLangConversion;
 345                 /* don't do anything if this is the conversion table */
 346                 if ( $parser->getTitle()->getNamespace() == NS_MEDIAWIKI &&
 347                                  strpos($parser->mTitle->getText(), "Conversiontable") !== false )
 348                 {
 349                         return $text;
 350                 }
 351
 352                 if($wgDisableLangConversion)
 353                         return $text;
 354
 355                 $text = $this->convert( $text );
 356                 $parser->mOutput->setTitleText( $this->mTitleDisplay );
 357                 return $text;
 358         }
 359
 360         /**
 361          *  convert title
 362          * @private
 363          */
 364         function convertTitle($text){
 365                 // check for __NOTC__ tag
 366                 if( !$this->mDoTitleConvert ) {
 367                         $this->mTitleDisplay = $text;
 368                         return $text;
 369                 }
 370
 371                 // use the title from the T flag if any
 372                 if($this->mTitleFromFlag){
 373                         $this->mTitleFromFlag = false;
 374                         return $this->mTitleDisplay;
 375                 }
 376
 377                 global $wgRequest;
 378                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 379                 $action = $wgRequest->getText( 'action' );
 380                 if ( $isredir == 'no' || $action == 'edit' ) {
 381                         return $text;
 382                 } else {
 383                         $this->mTitleDisplay = $this->convert($text);
 384                         return $this->mTitleDisplay;
 385                 }
 386         }
 387
 388         /**
 389          * convert text to different variants of a language. the automatic
 390          * conversion is done in autoConvert(). here we parse the text
 391          * marked with -{}-, which specifies special conversions of the
 392          * text that can not be accomplished in autoConvert()
 393          *
 394          * syntax of the markup:
 395          * -{code1:text1;code2:text2;...}-  or
 396          * -{flags|code1:text1;code2:text2;...}-  or
 397          * -{text}- in which case no conversion should take place for text
 398          *
 399          * @param string $text text to be converted
 400          * @param bool $isTitle whether this conversion is for the article title
 401          * @return string converted text
 402          * @public
 403          */
 404         function convert( $text , $isTitle=false) {
 405
 406                 $mw =& MagicWord::get( 'notitleconvert' );
 407                 if( $mw->matchAndRemove( $text ) )
 408                         $this->mDoTitleConvert = false;
 409                 $mw =& MagicWord::get( 'nocontentconvert' );
 410                 if( $mw->matchAndRemove( $text ) ) {
 411                         $this->mDoContentConvert = false;
 412                 }
 413
 414                 // no conversion if redirecting
 415                 $mw =& MagicWord::get( 'redirect' );
 416                 if( $mw->matchStart( $text ))
 417                         return $text;
 418
 419                 // for title convertion
 420                 if ($isTitle) return $this->convertTitle($text);
 421
 422                 $plang = $this->getPreferredVariant();
 423                 $tarray = explode($this->mMarkup['end'], $text);
 424                 $text = '';
 425                 foreach($tarray as $txt) {
 426                         $marked = explode($this->mMarkup['begin'], $txt, 2);
 427
 428                         if( $this->mDoContentConvert )
 429                                 $text .= $this->autoConvert($marked[0],$plang);
 430                         else
 431                                 $text .= $marked[0];
 432
 433                         if(array_key_exists(1, $marked)){
 434                                 // strip the flags from syntax like -{T| ... }-
 435                                 $crule = new ConverterRule($marked[1], $this);
 436                                 $crule->parse($plang);
 437
 438                                 $text .= $crule->getDisplay();
 439                                 $this->applyManualConv($crule);
 440                         }
 441                 }
 442
 443                 return $text;
 444         }
 445
 446         /**
 447          * if a language supports multiple variants, it is
 448          * possible that non-existing link in one variant
 449          * actually exists in another variant. this function
 450          * tries to find it. See e.g. LanguageZh.php
 451          *
 452          * @param string $link the name of the link
 453          * @param mixed $nt the title object of the link
 454          * @return null the input parameters may be modified upon return
 455          * @public
 456          */
 457         function findVariantLink( &$link, &$nt ) {
 458                 global $wgDisableLangConversion;
 459                 $linkBatch = new LinkBatch();
 460
 461                 $ns=NS_MAIN;
 462
 463                 if(is_object($nt))
 464                         $ns = $nt->getNamespace();
 465
 466                 $variants = $this->autoConvertToAllVariants($link);
 467                 if($variants == false) //give up
 468                         return;
 469
 470                 $titles = array();
 471
 472                 foreach( $variants as $v ) {
 473                         if($v != $link){
 474                                 $varnt = Title::newFromText( $v, $ns );
 475                                 if(!is_null($varnt)){
 476                                         $linkBatch->addObj($varnt);
 477                                         $titles[]=$varnt;
 478                                 }
 479                         }
 480                 }
 481
 482                 // fetch all variants in single query
 483                 $linkBatch->execute();
 484
 485                 foreach( $titles as $varnt ) {
 486                         if( $varnt->getArticleID() > 0 ) {
 487                                 $nt = $varnt;
 488                                 if( !$wgDisableLangConversion )
 489                                         $link = $v;
 490                                 break;
 491                         }
 492                 }
 493         }
 494
 495     /**
 496          * returns language specific hash options
 497          *
 498          * @public
 499          */
 500         function getExtraHashOptions() {
 501                 $variant = $this->getPreferredVariant();
 502                 return '!' . $variant ;
 503         }
 504
 505     /**
 506          * get title text as defined in the body of the article text
 507          *
 508          * @public
 509          */
 510         function getParsedTitle() {
 511                 return $this->mTitleDisplay;
 512         }
 513
 514         /**
 515          * a write lock to the cache
 516          *
 517          * @private
 518          */
 519         function lockCache() {
 520                 global $wgMemc;
 521                 $success = false;
 522                 for($i=0; $i<30; $i++) {
 523                         if($success = $wgMemc->add($this->mCacheKey . "lock", 1, 10))
 524                                 break;
 525                         sleep(1);
 526                 }
 527                 return $success;
 528         }
 529
 530         /**
 531          * unlock cache
 532          *
 533          * @private
 534          */
 535         function unlockCache() {
 536                 global $wgMemc;
 537                 $wgMemc->delete($this->mCacheKey . "lock");
 538         }
 539
 540
 541         /**
 542          * Load default conversion tables
 543          * This method must be implemented in derived class
 544          *
 545          * @private
 546          */
 547         function loadDefaultTables() {
 548                 $name = get_class($this);
 549                 wfDie("Must implement loadDefaultTables() method in class $name");
 550         }
 551
 552         /**
 553          * load conversion tables either from the cache or the disk
 554          * @private
 555          */
 556         function loadTables($fromcache=true) {
 557                 global $wgMemc;
 558                 if( $this->mTablesLoaded )
 559                         return;
 560                 wfProfileIn( __METHOD__ );
 561                 $this->mTablesLoaded = true;
 562                 $this->mTables = false;
 563                 if($fromcache) {
 564                         wfProfileIn( __METHOD__.'-cache' );
 565                         $this->mTables = $wgMemc->get( $this->mCacheKey );
 566                         wfProfileOut( __METHOD__.'-cache' );
 567                 }
 568                 if ( !$this->mTables || !isset( $this->mTables[self::CACHE_VERSION_KEY] ) ) {
 569                         wfProfileIn( __METHOD__.'-recache' );
 570                         // not in cache, or we need a fresh reload.
 571                         // we will first load the default tables
 572                         // then update them using things in MediaWiki:Zhconversiontable/*
 573                         $this->loadDefaultTables();
 574                         foreach($this->mVariants as $var) {
 575                                 $cached = $this->parseCachedTable($var);
 576                                 $this->mTables[$var]->mergeArray($cached);
 577                         }
 578
 579                         $this->postLoadTables();
 580                         $this->mTables[self::CACHE_VERSION_KEY] = true;
 581
 582                         if($this->lockCache()) {
 583                                 $wgMemc->set($this->mCacheKey, $this->mTables, 43200);
 584                                 $this->unlockCache();
 585                         }
 586                         wfProfileOut( __METHOD__.'-recache' );
 587                 }
 588                 wfProfileOut( __METHOD__ );
 589         }
 590
 591     /**
 592          * Hook for post processig after conversion tables are loaded
 593          *
 594          */
 595         function postLoadTables() {}
 596
 597     /**
 598          * Reload the conversion tables
 599          *
 600          * @private
 601          */
 602         function reloadTables() {
 603                 if($this->mTables)
 604                         unset($this->mTables);
 605                 $this->mTablesLoaded = false;
 606                 $this->loadTables(false);
 607         }
 608
 609
 610         /**
 611          * parse the conversion table stored in the cache
 612          *
 613          * the tables should be in blocks of the following form:
 614          *              -{
 615          *                      word => word ;
 616          *                      word => word ;
 617          *                      ...
 618          *              }-
 619          *
 620          *      to make the tables more manageable, subpages are allowed
 621          *      and will be parsed recursively if $recursive=true
 622          *
 623          */
 624         function parseCachedTable($code, $subpage='', $recursive=true) {
 625                 global $wgMessageCache;
 626                 static $parsed = array();
 627
 628                 if(!is_object($wgMessageCache))
 629                         return array();
 630
 631                 $key = 'Conversiontable/'.$code;
 632                 if($subpage)
 633                         $key .= '/' . $subpage;
 634
 635                 if(array_key_exists($key, $parsed))
 636                         return array();
 637
 638                 if ( strpos( $code, '/' ) === false ) {
 639                         $txt = $wgMessageCache->get( 'Conversiontable', true, $code );
 640                 } else {
 641                         $title = Title::makeTitleSafe( NS_MEDIAWIKI, "Conversiontable/$code" );
 642                         if ( $title && $title->exists() ) {
 643                                 $article = new Article( $title );
 644                                 $txt = $article->getContents();
 645                         } else {
 646                                 $txt = '';
 647                         }
 648                 }
 649
 650                 // get all subpage links of the form
 651                 // [[MediaWiki:conversiontable/zh-xx/...|...]]
 652                 $linkhead = $this->mLangObj->getNsText(NS_MEDIAWIKI) . ':Conversiontable';
 653                 $subs = explode('[[', $txt);
 654                 $sublinks = array();
 655                 foreach( $subs as $sub ) {
 656                         $link = explode(']]', $sub, 2);
 657                         if(count($link) != 2)
 658                                 continue;
 659                         $b = explode('|', $link[0]);
 660                         $b = explode('/', trim($b[0]), 3);
 661                         if(count($b)==3)
 662                                 $sublink = $b[2];
 663                         else
 664                                 $sublink = '';
 665
 666                         if($b[0] == $linkhead && $b[1] == $code) {
 667                                 $sublinks[] = $sublink;
 668                         }
 669                 }
 670
 671
 672                 // parse the mappings in this page
 673                 $blocks = explode($this->mMarkup['begin'], $txt);
 674                 array_shift($blocks);
 675                 $ret = array();
 676                 foreach($blocks as $block) {
 677                         $mappings = explode($this->mMarkup['end'], $block, 2);
 678                         $stripped = str_replace(array("'", '"', '*','#'), '', $mappings[0]);
 679                         $table = explode( ';', $stripped );
 680                         foreach( $table as $t ) {
 681                                 $m = explode( '=>', $t );
 682                                 if( count( $m ) != 2)
 683                                         continue;
 684                                 // trim any trailling comments starting with '//'
 685                                 $tt = explode('//', $m[1], 2);
 686                                 $ret[trim($m[0])] = trim($tt[0]);
 687                         }
 688                 }
 689                 $parsed[$key] = true;
 690
 691
 692                 // recursively parse the subpages
 693                 if($recursive) {
 694                         foreach($sublinks as $link) {
 695                                 $s = $this->parseCachedTable($code, $link, $recursive);
 696                                 $ret = array_merge($ret, $s);
 697                         }
 698                 }
 699
 700                 if ($this->mUcfirst) {
 701                         foreach ($ret as $k => $v) {
 702                                 $ret[Language::ucfirst($k)] = Language::ucfirst($v);
 703                         }
 704                 }
 705                 return $ret;
 706         }
 707
 708         /**
 709          * Enclose a string with the "no conversion" tag. This is used by
 710          * various functions in the Parser
 711          *
 712          * @param string $text text to be tagged for no conversion
 713          * @return string the tagged text
 714          * @public
 715          */
 716         function markNoConversion($text, $noParse=false) {
 717                 # don't mark if already marked
 718                 if(strpos($text, $this->mMarkup['begin']) ||
 719                    strpos($text, $this->mMarkup['end']))
 720                         return $text;
 721
 722                 $ret = $this->mMarkup['begin'] .'R|'. $text . $this->mMarkup['end'];
 723                 return $ret;
 724         }
 725
 726         /**
 727          * convert the sorting key for category links. this should make different
 728          * keys that are variants of each other map to the same key
 729          */
 730         function convertCategoryKey( $key ) {
 731                 return $key;
 732         }
 733         /**
 734          * hook to refresh the cache of conversion tables when
 735          * MediaWiki:conversiontable* is updated
 736          * @private
 737          */
 738         function OnArticleSaveComplete($article, $user, $text, $summary, $isminor, $iswatch, $section, $flags, $revision) {
 739                 $titleobj = $article->getTitle();
 740                 if($titleobj->getNamespace() == NS_MEDIAWIKI) {
 741                         $title = $titleobj->getDBkey();
 742                         $t = explode('/', $title, 3);
 743                         $c = count($t);
 744                         if( $c > 1 && $t[0] == 'Conversiontable' ) {
 745                                 if(in_array($t[1], $this->mVariants)) {
 746                                         $this->reloadTables();
 747                                 }
 748                         }
 749                 }
 750                 return true;
 751         }
 752
 753         /**
 754          * Armour rendered math against conversion
 755          * Wrap math into rawoutput -{R| math }- syntax
 756          * @public
 757          */
 758         function armourMath($text){
 759                 $ret = $this->mMarkup['begin'] . 'R|' . $text . $this->mMarkup['end'];
 760                 return $ret;
 761         }
 762 }
 763
 764 /**
 765  * parser for rules of language conversion , parse rules in -{ }- tag
 766  * @ingroup Language
 767  * @author  fdcn <fdcn64@gmail.com>
 768  */
 769 class ConverterRule {
 770         var $mText; // original text in -{text}-
 771         var $mConverter; // LanguageConverter object
 772         var $mManualCodeError='<strong class="error">code error!</strong>';
 773         var $mRuleDisplay = '',$mRuleTitle=false;
 774         var $mRules = '';// string : the text of the rules
 775         var $mRulesAction = 'none';
 776         var $mFlags = array();
 777         var $mConvTable = array();
 778         var $mBidtable = array();// array of the translation in each variant
 779         var $mUnidtable = array();// array of the translation in each variant
 780
 781         /**
 782          * Constructor
 783          *
 784          * @param string $text the text between -{ and }-
 785          * @param object $converter a  LanguageConverter object
 786          * @access public
 787          */
 788         function __construct($text,$converter){
 789                 $this->mText = $text;
 790                 $this->mConverter=$converter;
 791                 foreach($converter->mVariants as $v){
 792                         $this->mConvTable[$v]=array();
 793                 }
 794         }
 795
 796         /**
 797          * check if variants array in convert array
 798          *
 799          * @param string $variant Variant language code
 800          * @return string Translated text
 801          * @public
 802          */
 803         function getTextInBidtable($variants){
 804                 if(is_string($variants)){ $variants=array($variants); }
 805                 if(!is_array($variants)) return false;
 806                 foreach ($variants as $variant){
 807                         if(array_key_exists($variant, $this->mBidtable)){
 808                                 return $this->mBidtable[$variant];
 809                         }
 810                 }
 811                 return false;
 812         }
 813
 814         /**
 815          * Parse flags with syntax -{FLAG| ... }-
 816          * @private
 817          */
 818         function parseFlags(){
 819                 $text = $this->mText;
 820                 if(strlen($text) < 2 ) {
 821                         $this->mFlags = array( 'R' );
 822                         $this->mRules = $text;
 823                         return;
 824                 }
 825
 826                 $flags = array();
 827                 $markup = $this->mConverter->mMarkup;
 828                 $validFlags = $this->mConverter->mFlags;
 829
 830                 $tt = explode($markup['flagsep'], $text, 2);
 831                 if(count($tt) == 2) {
 832                         $f = explode($markup['varsep'], $tt[0]);
 833                         foreach($f as $ff) {
 834                                 $ff = trim($ff);
 835                                 if(array_key_exists($ff, $validFlags) &&
 836                                                         !in_array($validFlags[$ff], $flags))
 837                                         $flags[] = $validFlags[$ff];
 838                         }
 839                         $rules = $tt[1];
 840                 } else {
 841                         $rules = $text;
 842                 }
 843
 844                 //check flags
 845                 if( in_array('R',$flags) ){
 846                         $flags = array('R');// remove other flags
 847                 } elseif ( in_array('N',$flags) ){
 848                         $flags = array('N');// remove other flags
 849                 } elseif ( in_array('-',$flags) ){
 850                         $flags = array('-');// remove other flags
 851                 } elseif (count($flags)==1 && $flags[0]=='T'){
 852                         $flags[]='H';
 853                 } elseif ( in_array('H',$flags) ){
 854                         // replace A flag, and remove other flags except T
 855                         $temp=array('+','H');
 856                         if(in_array('T',$flags)) $temp[] = 'T';
 857                         if(in_array('D',$flags)) $temp[] = 'D';
 858                         $flags = $temp;
 859                 } else {
 860                         if ( in_array('A',$flags)) {
 861                                 $flags[]='+';
 862                                 $flags[]='S';
 863                         }
 864                         if ( in_array('D',$flags) )
 865                                 $flags=array_diff($flags,array('S'));
 866                 }
 867                 if ( count($flags)==0 )
 868                         $flags = array('S');
 869                 $this->mRules=$rules;
 870                 $this->mFlags=$flags;
 871         }
 872
 873         /**
 874          * generate conversion table
 875          * @private
 876          */
 877         function parseRules() {
 878                 $rules = $this->mRules;
 879                 $flags = $this->mFlags;
 880                 $bidtable = array();
 881                 $unidtable = array();
 882                 $markup = $this->mConverter->mMarkup;
 883
 884                 $choice = explode($markup['varsep'], $rules );
 885                 foreach($choice as $c) {
 886                         $v = explode($markup['codesep'], $c);
 887                         if(count($v) != 2)
 888                                 continue;// syntax error, skip
 889                         $to=trim($v[1]);
 890                         $v=trim($v[0]);
 891                         $u = explode($markup['unidsep'], $v);
 892                         if(count($u) == 1) {
 893                                 $bidtable[$v] = $to;
 894                         } else if(count($u) == 2){
 895                                 $from=trim($u[0]);$v=trim($u[1]);
 896                                 if( array_key_exists($v,$unidtable) && !is_array($unidtable[$v]) )
 897                                         $unidtable[$v]=array($from=>$to);
 898                                 else
 899                                         $unidtable[$v][$from]=$to;
 900                         }
 901                         // syntax error, pass
 902                         if (!array_key_exists($v,$this->mConverter->mVariantNames)){
 903                                 $bidtable = array();
 904                                 $unidtable = array();
 905                                 break;
 906                         }
 907                 }
 908                 $this->mBidtable = $bidtable;
 909                 $this->mUnidtable = $unidtable;
 910         }
 911
 912         /**
 913          * @private
 914          */
 915         function getRulesDesc(){
 916                 $codesep = $this->mConverter->mDescCodeSep;
 917                 $varsep = $this->mConverter->mDescVarSep;
 918                 $text='';
 919                 foreach($this->mBidtable as $k => $v)
 920                         $text .= $this->mConverter->mVariantNames[$k]."$codesep$v$varsep";
 921                 foreach($this->mUnidtable as $k => $a)
 922                         foreach($a as $from=>$to)
 923                                 $text.=$from.'⇒'.$this->mConverter->mVariantNames[$k]."$codesep$to$varsep";
 924                 return $text;
 925         }
 926
 927         /**
 928          *  Parse rules conversion
 929          * @private
 930          */
 931         function getRuleConvertedStr($variant,$doConvert){
 932                 $bidtable = $this->mBidtable;
 933                 $unidtable = $this->mUnidtable;
 934
 935                 if( count($bidtable) + count($unidtable) == 0 ){
 936                         return $this->mRules;
 937                 } elseif ($doConvert){// the text converted
 938                         // display current variant in bidirectional array
 939                         $disp = $this->getTextInBidtable($variant);
 940                         // or display current variant in fallbacks
 941                         if(!$disp)
 942                                 $disp = $this->getTextInBidtable(
 943                                                 $this->mConverter->getVariantFallbacks($variant));
 944                         // or display current variant in unidirectional array
 945                         if(!$disp && array_key_exists($variant,$unidtable)){
 946                                 $disp = array_values($unidtable[$variant]);
 947                                 $disp = $disp[0];
 948                         }
 949                         // or display frist text under disable manual convert
 950                         if(!$disp && $this->mConverter->mManualLevel[$variant]=='disable') {
 951                                 if(count($bidtable)>0){
 952                                         $disp = array_values($bidtable);
 953                                         $disp = $disp[0];
 954                                 } else {
 955                                         $disp = array_values($unidtable);
 956                                         $disp = array_values($disp[0]);
 957                                         $disp = $disp[0];
 958                                 }
 959                         }
 960                         return $disp;
 961                 } else {// no convert
 962                         return $this->mRules;
 963                 }
 964         }
 965
 966         /**
 967          * generate conversion table for all text
 968          * @private
 969          */
 970         function generateConvTable(){
 971                 $flags = $this->mFlags;
 972                 $bidtable = $this->mBidtable;
 973                 $unidtable = $this->mUnidtable;
 974                 $manLevel = $this->mConverter->mManualLevel;
 975
 976                 $vmarked=array();
 977                 foreach($this->mConverter->mVariants as $v) {
 978                         /* for bidirectional array
 979                                 fill in the missing variants, if any,
 980                                 with fallbacks */
 981                         if(!array_key_exists($v, $bidtable)) {
 982                                 $variantFallbacks = $this->mConverter->getVariantFallbacks($v);
 983                                 $vf = $this->getTextInBidtable($variantFallbacks);
 984                                 if($vf) $bidtable[$v] = $vf;
 985                         }
 986
 987                         if(array_key_exists($v,$bidtable)){
 988                                 foreach($vmarked as $vo){
 989                                         // use syntax: -{A|zh:WordZh;zh-tw:WordTw}-
 990                                         // or -{H|zh:WordZh;zh-tw:WordTw}- or -{-|zh:WordZh;zh-tw:WordTw}-
 991                                         // to introduce a custom mapping between
 992                                         // words WordZh and WordTw in the whole text
 993                                         if($manLevel[$v]=='bidirectional'){
 994                                                 $this->mConvTable[$v][$bidtable[$vo]]=$bidtable[$v];
 995                                         }
 996                                         if($manLevel[$vo]=='bidirectional'){
 997                                                 $this->mConvTable[$vo][$bidtable[$v]]=$bidtable[$vo];
 998                                         }
 999                                 }
1000                                 $vmarked[]=$v;
1001                         }
1002                         /*for unidirectional array
1003                                 fill to convert tables */
1004                         $allow_unid = $manLevel[$v]=='bidirectional'
1005                                         || $manLevel[$v]=='unidirectional';
1006                         if($allow_unid && array_key_exists($v,$unidtable)){
1007                                 $ct=$this->mConvTable[$v];
1008                                 $this->mConvTable[$v] = array_merge($ct,$unidtable[$v]);
1009                         }
1010                 }
1011         }
1012
1013         /**
1014          * Parse rules and flags
1015          * @public
1016          */
1017         function parse($variant){
1018                 if(!$variant) $variant = $this->mConverter->getPreferredVariant();
1019
1020                 $this->parseFlags();
1021                 $flags = $this->mFlags;
1022
1023                 if( !in_array('R',$flags) || !in_array('N',$flags) ){
1024                         //FIXME: may cause trouble here...
1025                         //strip &nbsp; since it interferes with the parsing, plus,
1026                         //all spaces should be stripped in this tag anyway.
1027                         $this->mRules = str_replace('&nbsp;', '', $this->mRules);
1028                         // decode => HTML entities modified by Sanitizer::removeHTMLtags
1029                         $this->mRules = str_replace('=&gt;','=>',$this->mRules);
1030
1031                         $this->parseRules();
1032                 }
1033                 $rules = $this->mRules;
1034
1035                 if(count($this->mBidtable)==0 && count($this->mUnidtable)==0){
1036                         if(in_array('+',$flags) || in_array('-',$flags))
1037                                 // fill all variants if text in -{A/H/-|text} without rules
1038                                 foreach($this->mConverter->mVariants as $v)
1039                                         $this->mBidtable[$v] = $rules;
1040                         elseif (!in_array('N',$flags) && !in_array('T',$flags) )
1041                                 $this->mFlags = $flags = array('R');
1042                 }
1043
1044                 if( in_array('R',$flags) ) {
1045                         // if we don't do content convert, still strip the -{}- tags
1046                         $this->mRuleDisplay = $rules;
1047                 } elseif ( in_array('N',$flags) ){
1048                         // proces N flag: output current variant name
1049                         $this->mRuleDisplay = $this->mConverter->mVariantNames[trim($rules)];
1050                 } elseif ( in_array('D',$flags) ){
1051                         // proces D flag: output rules description
1052                         $this->mRuleDisplay = $this->getRulesDesc();
1053                 } elseif ( in_array('H',$flags) || in_array('-',$flags) ) {
1054                         // proces H,- flag or T only: output nothing
1055                         $this->mRuleDisplay = '';
1056                 } elseif ( in_array('S',$flags) ){
1057                         $this->mRuleDisplay = $this->getRuleConvertedStr($variant,
1058                                                         $this->mConverter->mDoContentConvert);
1059                 } else {
1060                         $this->mRuleDisplay= $this->mManualCodeError;
1061                 }
1062                 // proces T flag
1063                 if ( in_array('T',$flags) ) {
1064                         $this->mRuleTitle = $this->getRuleConvertedStr($variant,
1065                                                         $this->mConverter->mDoTitleConvert);
1066                 }
1067
1068                 if (in_array('-', $flags))
1069                         $this->mRulesAction='remove';
1070                 if (in_array('+', $flags))
1071                         $this->mRulesAction='add';
1072
1073                 $this->generateConvTable();
1074         }
1075
1076         /**
1077          * @public
1078          */
1079         function hasRules(){
1080                 // TODO:
1081         }
1082
1083         /**
1084          * get display text on markup -{...}-
1085          * @public
1086          */
1087         function getDisplay(){
1088                 return $this->mRuleDisplay;
1089         }
1090         /**
1091          * get converted title
1092          * @public
1093          */
1094         function getTitle(){
1095                 return $this->mRuleTitle;
1096         }
1097
1098         /**
1099          * return how deal with conversion rules
1100          * @public
1101          */
1102         function getRulesAction(){
1103                 return $this->mRulesAction;
1104         }
1105
1106         /**
1107          * get conversion table ( bidirectional and unidirectional conversion table )
1108          * @public
1109          */
1110         function getConvTable(){
1111                 return $this->mConvTable;
1112         }
1113
1114         /**
1115          * get conversion rules string
1116          * @public
1117          */
1118         function getRules(){
1119                 return $this->mRules;
1120         }
1121
1122         /**
1123          * get conversion flags
1124          * @public
1125          */
1126         function getFlags(){
1127                 return $this->mFlags;
1128         }
1129 }